In the past, we had many discussions about bit depth and we had the luck that knowledgable people Like Nika Aldrich and Bob Katz came here to explain what digital recording was all about. They tackled all of the myths that are floating around about bit depth. Steinberg had a hard time explaining that 32bit floating Point was as good, if not better- than the 48bit fixed Point that PT used back then.
Even today, there is much ignorance and a lot of people think that “resolution” in audio means the same thing as “resolution” in video. Just yesterday I had a long discussionwith video people who insisted that the audio department worked in 24bit, because they would not take any compromise in terms of audio quality.
Anyway, I just came across this article; hope it is of some help.
Thanks Fredo. I think we just need to keep repeating it for people to believe. It’s like the old war in CPUs that more GHz is better. Today, this law doesn’t work anymore as there are more cores and better parallel programming. Or that more Megapixels in a camera are better (while the chip size stays the same and the pixels get smaller and smaller, producing a lot more noise, making the image worse). Thanks for that article and reminder.
Well, for once Fredo, I don’t completely agree.
I just read that interesting thing about bits…
It is kind of true, yes… But only because it mainly talks about music applications, where in fact, with absolutely no dynamic in the final mix, there is little difference between 32float and 24 bits. (even 16, indeed)
But it does’ change the fact that most daws audio engines run now at 32float, no matter what you do.
Even if you ask it to express the result at 24 bits, it does’ change the fact that all calculations remain at the same, native daw audio engine resolution.
So why would we want to work at anything but the non-modifiable daw precision as long as the main mix is not finished ?
Why loose the precision as long as we have some more things to do ? (with off-line processes for instance, or bounce… every time we write a new file before reaching the end of the mix.)
At the end, right, we end up with a 24 bits mantisse and a 8 bit exponent.
Exponent which is now useless, as we don’t have any more calculations to do…
So we can now safely drop them and express the last result of our calculations in 24bits. (final mix, in a word…)
This is then really coherent with the fact that we use 24bits for recording on the field, and 24bits for deliverables.
But 32bits is the way to go to keep the precision up during all editing and mixing work in my opinion…
Not to say that it’s elegantly adapted with 64 bits accumulators and 64bits OS, Hosts and SDK architectures…
So at the end of the day, I do think that for all those reasons, it’s way more logical to work everything in 32float and deliver the result at the very end in 24bits.
(or shoot at 4k, and deliver a dip at 2k, which is enough good when coloring and vfx are done.)
Just my 2 cents (and a post-production only view, that right
Haven’t read it tbh, but he probably misses sound design (most people do).
When you pitch shift stuff (down) frequencies that were inaudible suddenly appear in the conceivable parts of the audio spectrum when you record stuff at higher sample rates.
Oliver, yes it does mention this. And this is a valid usage of higher sample rates. But of course you need the gear. Not only do you need the recording hardware set to 192kHz, but also a mic that actually is able to record those frequencies. There are a couple of those, Sanken (SANKEN MICROPHONE CO .,LTD. | Product [ CO-100K ]) for example up to 100kHz. The latest Godzilla movie used a lot of this (Soundworks Collection).
I think the articles go specifically not towards “recording” but “listening”. Listening to or paying more for an apparently superior sound quality music song is useless. There’s this Pono Player movement from Neil Young for example that claims superior sound quality only because of “high resolution” audio. And for that purpose, I think we agree, high resolution is bogus. See and go to minute 6:11. Mind boggling. https://www.kickstarter.com/projects/1003614822/ponomusic-where-your-soul-rediscovers-music/description
Also, some sound fx outlets do advertise their cool 192kHz QUALITY samples. But when you look at the recording in a spectral view, it all tops off at 16-18kHz. Telling you they didn’t record it with the correct mic
Haven’t thought about that until now but from what I read up and learned about this matter I’d say it doesn’t matter. And as most of us probably work in 48kHz, 96kHz would be the way to go if you want divisible by 2.
Also outdated these days.
Most if not all of the really good SRC use the greatest Common Denominator mode, and only calculate in whole numbers by first upsampling and then going down.
I cannot remember the exact maths, but it works well in integer values.
There was a study recently though that showed people were aware of far higher frequencies than previously thought - I cant find it now. I was suprised myself - though I am in general agreement with you.
24-bit is better than 16-bits & no amount of straw man arguments from the quoted sonicscoop.com website will change this (8-bit digital effects are still better than cassette tapes, forsooth!)
Trouble is I have written this reply well over a dozen times & each time it has been deleted because it ended up, well - wandering a little from the initial path which would appear to be impossible to completely avoid. I’ve gone in and out of loudness wars (which are still well on, despite what we want to believe to the contrary) but at the end of the day it all boils down to that 24-bits are better than 16-bits and by a not insignificant amount either.
16-bit dither is an extremely low 12dB SPL. All but inaudible.
Moving to 24-bit. 24-bit dither would be at an inaudible -36dB SPL. The DAC noise at -17dB SPL is well below the threshold of hearing. The difficult part always seeming to need lots of explanation is how a 24-bit system really being 48dB better than a 16-bit system because a 24-bit recording would have to be lowered in level by 48dB in order to reduce it to the SNR of 16-bit, as this takes so many necessary but lengthy detours.
But the numbers do not lie - and I am not talking about magical marketing bits here but real, achievable ones.
24-bit is 48dB better than 16-bit so the video department are perfectly correct to insist on this, especially when they also must now comply with loudness specifications.
From my understanding (I’m a sound designer) it’s not so much about capturing content above the upper threshold of human hearing as it is about having additional samples to play with when manipulating content. If you halve the playback fequency of a 48KHz sample, you’re effectively resampling to 24KHz. Content with higher sample rates tends to be more robust in these kinds of scenarios.
Same goes for bit depth when you’re making radical alterations to dynamics.
Any critical process is executed within double precision (64bit). That was so within the old engine, and is still so in the new.
Plugins handling critical processes also upsample to double precision.
Old VST2 standard was 32bit in- and out.
VST 3 is 64bit in- and out.
Making the engine full double precision eliminates the need for upsampling and truncating before and after each insert slot and from each channel to bus/bus/output. So since the plugins are (or can be) 64bit nowadays, there is no reason at all for “converting” before and after the inster slots.
So what happens is that by having a 64 bit engine from start to end, a lot of unnessesairy processing is removed from the audio engine.
Does this make a difference in sound quality: Nope. Only under exotic laboratory conditions you would be able to expose the “gain in quality”.
It does simplify the piping/processing and programming throughout the audio engine?
But indeed, the majority of people still thinks that “more is better”, so even for that reason alone, the change is justified.